504 lines
13 KiB
YAML
504 lines
13 KiB
YAML
# tr.yaml
|
||
# -------
|
||
# Turkish language specification
|
||
|
||
components:
|
||
level:
|
||
null_probability: 0.9
|
||
alphanumeric_probability: 0.1
|
||
|
||
staircase:
|
||
null_probability: 0.99
|
||
alphanumeric_probability: 0.01
|
||
|
||
entrance:
|
||
null_probability: 0.999
|
||
alphanumeric_probability: 0.001
|
||
|
||
unit:
|
||
null_probability: 0.7
|
||
alphanumeric_probability: 0.3
|
||
|
||
combinations:
|
||
-
|
||
components:
|
||
- house_number
|
||
- staircase
|
||
- level
|
||
- unit
|
||
label: house_number
|
||
separators:
|
||
- separator: "/"
|
||
probability: 0.95
|
||
- separator: "-"
|
||
probability: 0.05
|
||
probability: 0.005
|
||
-
|
||
components:
|
||
- house_number
|
||
- level
|
||
- unit
|
||
label: house_number
|
||
separators:
|
||
- separator: "/"
|
||
probability: 0.95
|
||
- separator: "-"
|
||
probability: 0.05
|
||
probability: 0.005
|
||
-
|
||
components:
|
||
- house_number
|
||
- level
|
||
label: house_number
|
||
separators:
|
||
- separator: "/"
|
||
probability: 0.95
|
||
- separator: "-"
|
||
probability: 0.05
|
||
probability: 0.1
|
||
# For unit types like 2/34
|
||
-
|
||
components:
|
||
- house_number
|
||
- unit
|
||
label: house_number
|
||
separators:
|
||
- separator: "/"
|
||
probability: 0.95
|
||
- separator: "-"
|
||
probability: 0.05
|
||
probability: 0.005
|
||
|
||
|
||
numbers:
|
||
|
||
default: &numara
|
||
canonical: numara
|
||
abbreviated: "no:"
|
||
sample: true
|
||
canonical_probability: 0.3
|
||
abbreviated_probability: 0.6
|
||
sample_probability: 0.1
|
||
numeric:
|
||
direction: left
|
||
numeric_affix:
|
||
affix: "no:"
|
||
whitespace_probability: 0.6
|
||
direction: left
|
||
numeric_probability: 0.4
|
||
numeric_affix_probability: 0.6
|
||
|
||
alphanumeric_phrase_probability: 0.05
|
||
no_number_probability: 0.05
|
||
|
||
|
||
and:
|
||
default: &ve
|
||
canonical: ve
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
|
||
|
||
cross_streets:
|
||
ve: *ve
|
||
corner_of: &kose
|
||
canonical: köşe
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
kosesinde: &kosesinde
|
||
canonical: köşesinde
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
intersection:
|
||
default: *ve
|
||
probability: 0.8
|
||
alternatives:
|
||
- alternative: *kose
|
||
probability: 0.1
|
||
- alternative: *kosesinde
|
||
probability: 0.1
|
||
|
||
arasinda: &arasinda
|
||
canonical: arasında
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
parentheses_probability: 0.5
|
||
between:
|
||
default: *arasinda
|
||
|
||
levels:
|
||
kat: &kat
|
||
canonical: kat
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
direction_probability: 0.9
|
||
digits:
|
||
ascii_probability: 0.7
|
||
roman_numeral_probability: 0.3
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.1
|
||
ordinal:
|
||
direction: right
|
||
digits:
|
||
ascii_probability: 0.3
|
||
roman_numeral_probability: 0.7
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.1
|
||
numeric_probability: 0.4
|
||
ordinal_probability: 0.6
|
||
|
||
zemin_kat: &zemin_kat
|
||
canonical: zemin kat
|
||
abbreviated: zk
|
||
sample: true
|
||
canonical_probability: 0.3
|
||
abbreviated_probability: 0.4
|
||
sample_probability: 0.3
|
||
asma_kat: &asma_kat
|
||
canonical: asma kat
|
||
half_floors: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
sample: true
|
||
# e.g. asma kat 2
|
||
numeric:
|
||
direction: left
|
||
# e.g. 2. asma kat
|
||
ordinal:
|
||
direction: right
|
||
numeric_probability: 0.1
|
||
ordinal_probability: 0.2
|
||
standalone_probability: 0.6
|
||
bodrum: &bodrum
|
||
canonical: bodrum
|
||
sample: true
|
||
canonical_probability: 0.7
|
||
sample_probability: 0.3
|
||
# e.g. bodrum 1
|
||
numeric:
|
||
direction: left
|
||
direction_probability: 0.8
|
||
# e.g. 1. bodrum
|
||
ordinal:
|
||
direction: right
|
||
digits:
|
||
ascii_probability: 0.7
|
||
roman_numeral_probability: 0.3
|
||
standalone_probability: 0.99
|
||
number_abs_value: true
|
||
number_min_abs_value: 1
|
||
numeric_probability: 0.005
|
||
ordinal_probability: 0.005
|
||
|
||
aliases:
|
||
"<-1":
|
||
default: *bodrum
|
||
"-1":
|
||
default: *bodrum
|
||
# Special token for half-floors
|
||
half_floors:
|
||
default: *asma_kat
|
||
"0":
|
||
default: *zemin_kat
|
||
probability: 0.9
|
||
alternatives:
|
||
- alternative: *kat
|
||
probability: 0.1
|
||
|
||
numbering_starts_at: 0
|
||
|
||
alphanumeric:
|
||
default: *kat
|
||
numeric_probability: 0.99 # With this probability, pick an integer
|
||
alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
|
||
numeric_plus_alpha_probability: 0.0001 # e.g. 2A
|
||
alpha_plus_numeric_probability: 0.0001 # e.g. A2
|
||
|
||
|
||
directions:
|
||
right: &sag
|
||
canonical: sağ
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: right
|
||
left: &sol
|
||
canonical: sol
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: right
|
||
alternatives:
|
||
- alternative: *sag
|
||
probability: 0.5
|
||
- alternative: *sol
|
||
probability: 0.5
|
||
|
||
cardinal_directions:
|
||
east: &dogu
|
||
canonical: doğu
|
||
abbreviated: d
|
||
canonical_probability: 0.95
|
||
abbreviated_probability: 0.05
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: d
|
||
direction: right
|
||
numeric_probability: 0.5
|
||
numeric_affix_probability: 0.5
|
||
|
||
west: &bati
|
||
canonical: batı
|
||
abbreviated: b
|
||
canonical_probability: 0.95
|
||
abbreviated_probability: 0.05
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: b
|
||
direction: right
|
||
numeric_probability: 0.5
|
||
numeric_affix_probability: 0.5
|
||
|
||
north: &kuzey
|
||
canonical: kuzey
|
||
abbreviated: k
|
||
canonical_probability: 0.95
|
||
abbreviated_probability: 0.05
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: k
|
||
direction: right
|
||
numeric_probability: 0.5
|
||
numeric_affix_probability: 0.5
|
||
|
||
south: &guney
|
||
canonical: güney
|
||
abbreviated: g
|
||
sample: true
|
||
canonical_probability: 0.75
|
||
abbreviated_probability: 0.1
|
||
sample_probability: 0.15
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: g
|
||
direction: right
|
||
numeric_probability: 0.5
|
||
numeric_affix_probability: 0.5
|
||
|
||
alternatives:
|
||
- alternative: *kuzey
|
||
probability: 0.25
|
||
- alternative: *dogu
|
||
probability: 0.23
|
||
- alternative: *guney
|
||
probability: 0.23
|
||
- alternative: *bati
|
||
probability: 0.23
|
||
|
||
entrances:
|
||
giris: &giris
|
||
canonical: giriş
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
|
||
# giriş 1, giriş A, etc.
|
||
alphanumeric: &entrance_alphanumeric
|
||
default: *giris
|
||
numeric_probability: 0.1 # e.g. giriş 1
|
||
alpha_probability: 0.85 # e.g. giriş A
|
||
numeric_plus_alpha_probability: 0.025 # e.g. 1A
|
||
alpha_plus_numeric_probability: 0.025 # e.g. A1
|
||
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|
||
|
||
|
||
staircases:
|
||
merdiven: &merdiven
|
||
canonical: merdiven
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
|
||
|
||
alphanumeric: &staircase_alphanumeric
|
||
default: *merdiven
|
||
numeric_probability: 0.75
|
||
alpha_probability: 0.2
|
||
numeric_plus_alpha_probability: 0.025
|
||
alpha_plus_numeric_probability: 0.025
|
||
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|
||
|
||
directional:
|
||
direction: right
|
||
direction_probability: 0.85
|
||
modifier:
|
||
alternatives:
|
||
- alternative: *sag
|
||
probability: 0.2
|
||
- alternative: *sol
|
||
probability: 0.2
|
||
- alternative: *kuzey
|
||
probability: 0.15
|
||
- alternative: *guney
|
||
probability: 0.15
|
||
- alternative: *dogu
|
||
probability: 0.15
|
||
- alternative: *bati
|
||
probability: 0.15
|
||
|
||
po_boxes:
|
||
posta_kutusu: &posta_kutusu
|
||
canonical: posta kutusu
|
||
abbreviated: pk
|
||
sample: true
|
||
canonical_probability: 0.2
|
||
abbreviated_probability: 0.4
|
||
sample_probability: 0.4
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.2
|
||
|
||
alphanumeric:
|
||
default: *posta_kutusu
|
||
numeric_probability: 0.9 # pp 123
|
||
alpha_probability: 0.05 # p.p A
|
||
numeric_plus_alpha_probability: 0.04 # pp 123G
|
||
alpha_plus_numeric_probability: 0.01 # pp A123
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|
||
|
||
digits:
|
||
- length: 1
|
||
probability: 0.05
|
||
- length: 2
|
||
probability: 0.1
|
||
- length: 3
|
||
probability: 0.2
|
||
- length: 4
|
||
probability: 0.5
|
||
- length: 5
|
||
probability: 0.1
|
||
- length: 6
|
||
probability: 0.05
|
||
|
||
|
||
units:
|
||
daire: &daire
|
||
canonical: daire
|
||
abbreviated: d
|
||
sample: true
|
||
canonical_probability: 0.4
|
||
abbreviated_probability: 0.4
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.1
|
||
apartman: &apartman
|
||
canonical: apartman
|
||
abbreviated: apt
|
||
sample: true
|
||
canonical_probability: 0.4
|
||
abbreviated_probability: 0.2
|
||
sample_probability: 0.4
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.1
|
||
|
||
oda: &oda
|
||
canonical: oda
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.1
|
||
ofis: &ofis
|
||
canonical: ofis
|
||
sample: true
|
||
canonical_probability: 0.6
|
||
sample_probability: 0.4
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.1
|
||
|
||
alphanumeric: &unit_alphanumeric
|
||
default: *daire
|
||
probability: 0.6
|
||
alternatives:
|
||
- alternative: *apartman
|
||
probability: 0.3
|
||
- alternative: *oda
|
||
probability: 0.1
|
||
numeric_probability: 0.9 # e.g. d. 1
|
||
numeric_plus_alpha_probability: 0.03 # e.g. 1A
|
||
alpha_plus_numeric_probability: 0.03 # e.g. A1
|
||
alpha_probability: 0.04 # e.g. daire A
|
||
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|
||
|
||
# If there are 10 floors, create unit numbers like #301 or #1032
|
||
use_floor_probability: 0.05
|
||
|
||
zones:
|
||
commercial: &commercial_unit_types
|
||
default: *oda
|
||
probability: 0.6
|
||
alternatives:
|
||
- alternative: *ofis
|
||
probability: 0.4
|
||
numeric_probability: 0.95 # e.g. oda 1
|
||
numeric_plus_alpha_probability: 0.01 # e.g. oda 1A
|
||
alpha_plus_numeric_probability: 0.01 # e.g. oda A1
|
||
alpha_probability: 0.03 # e.g. oda A
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|
||
university:
|
||
default: *oda
|
||
numeric_probability: 0.95 # e.g. oda 1
|
||
numeric_plus_alpha_probability: 0.01 # e.g. oda 1A
|
||
alpha_plus_numeric_probability: 0.01 # e.g. oda A1
|
||
alpha_probability: 0.03 # e.g. oda A
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|